SOEE3260/5670M P4 - 1

Global Seismology
Practical 4: Surface waves

Objectives

• Familiarize yourself with Love and Rayleigh waves at teleseismic distances.

• Determine a dispersion curve for a variety of periods and compare it with expected
dispersion curves along similar paths.

Seismic surface waves: a brief introduction

Characteristics of surface waves

The largest amplitudes observed on seismograms from shallow earthquakes are usually due to
surface waves rather than body waves. Surface waves travel horizontally close to the surface of
the Earth along great-circle paths. Their displacement amplitudes generally decrease with
increasing depth. Deeper events are known to produce weaker surface waves (if any at all).
Hence, the observation of strong surface waves indicates a shallow source depth. Deviations
from the typical great-circle paths mentioned above may occur near strong lateral
inhomogeneities (in the crust or upper mantle); even lateral reflections at such inhomogeneities
have been observed (e.g. Ward et al., 2021). Surface waves may travel considerable distances
due to their slow energy loss. For shallow earthquakes of magnitude Mb = 6 (or bigger) long-
period surface waves of periods T > 90s that have encircled the globe several times are a
common observation.

For an isotropic layered subsurface (the Earth fulfils this condition fairly well) the velocities at
which surface waves travel depend on frequency. This phenomenon is called dispersion. Note
that dispersion is not observed for body waves in such media if we disregard anelasticity. Even
if we consider anelastic absorption and scattering in inhomogeneous media the corresponding
dispersion of body waves is usually much smaller than the typical dispersion associated with
surface waves.

There are two types of surface waves: Rayleigh waves and Love waves. Again, for an isotropic
layered subsurface Rayleigh waves propagate P-SV energy whereas Love waves propagate SH
energy. Rayleigh waves are polarized in a vertical plane (i.e. no transverse component). The
particle motion at the surface is retrograde elliptical, and the vertical component is usually
bigger than the radial component by a factor of roughly 1.5. Love waves are polarized
horizontally (perpendicular to the direction of propagation). Deviations from these idealized
polarization characteristics are observed in highly heterogeneous and/or anisotropic regions.

Why study surface waves?

Surface waves are of interest in seismology because:

• Surface waves carry useful information about the structure of the crust and upper mantle.
• Most of the damage to man-made structures caused by earthquakes is due to surface
waves and not by body waves.
SOEE3260/5670M P4 - 2

Data
To study a few characteristics of Rayleigh waves we will use a large, shallow earthquake in the
Vanuatu region in the south-west Pacific. It is a large dip-slip event related to the subduction
of the Pacific plate in this area. The location of the earthquake according to the USGS National
Earthquake Information Center is 16.393° S; 168.314° E, Depth: 24.3 km, origin time:
26/11/1999 - 13:21:15.0. You can find more detail on this earthquake e.g. on the site
http:///www.globalcmt.org. We use records from a station of the Global Seismic Network in
Cathedral Cave Missouri (CCM) with a location of 38.056°N; 91.245°W (see
http://ds.iris.edu/gmap/#network=IU&station=CCM&planet=earth for more detail). The data
are from the main instrument (an STS1) at location “00”.

Download from Minerva the tar-ball 26-NOV-1999_1321.tar.gz which contains the following
SeismicHandler (Stammler, 1993) data files:
26-NOV-1999_1321.QBN
26-NOV-1999_1321.QHD

This file contains the three components (Z,N,E) of the station CCM. The traces start at 26-
NOV-1999_13:30:32.00 and are 3600s long.

You might have to take the difference between the start time and the origin time into
account when you calculate the travel time of the surface waves.

Also download the SH script

OPEN_XH3.SHC
into the same directory.

Question/Task 1: Start SeismicHandler from the command line using shc and load the traces
into SeismicHandler using the command

As before (and remember what you learned before) you can use the following commands to
change amplitudes, normalization, time windows etc. within SeismicHandler

xopen : Open graphics window (try xopen;; to skip the interactive part)
read name all : Read data in file name (no extension)
zoom 5 3 : Increase display amplitude for trace number 5 by a factor of 3
markphas : mark arrival time for a set of predetermined phases
phc : print hard copy – gives PS file – output filename is being given.
del all : Erase all traces from screen and memory
hide 1-3 : hides traces one to three from the display (kept in memory)
display h:all : displays all traces from memory in graphics window
stw : set time window. Cursor driven selection of time window
stw 100 200 : set time window to 100 to 200 s of data
dtw : Delete time window
rd : redraw – Refresh display without doing anything else
SOEE3260/5670M P4 - 3

makef : Make filter. Works interactively through a set of options

fili f <filter>: defines (fourier) filter for filtering
filter f all : filters data using the defined (fourier) filter
write name all : write out data on display into file name
quit y : exit SH

For further information on these commands please use the built-in man pages of
SeismicHandler using the help command.
Try to identify the different phases shown in the record. You can mark the body waves using
the command markphas 1 0.

Analysis of surface waves

Question/Task 2: Calculate the distance and backazimuth between source and receiver with
the given coordinates. Having the location of the source as lat1, lon1 and of the receiver as lat2,
lon2 you can calculate the distance  in degrees between these points on a sphere with

∆[°] = cos −1(cos(90 − 𝑙𝑎𝑡2) ∙

∙ cos(90 − 𝑙𝑎𝑡1) + sin(90 − 𝑙𝑎𝑡1) ∙ sin(90 − 𝑙𝑎𝑡2) ∙ cos(𝑙𝑜𝑛2 − 𝑙𝑜𝑛1))

To convert this distance  [º] to [km] you can use that 1º at the Earth’s surface is equivalent
to 111.19 km. Use python (or any other means) to code up this simple equation, but please note
that the angles have to be given in degrees. The azimuth between two points can be calculated
similarly

𝜃 = sin−1(sin(90 − 𝑙𝑎𝑡2) ∙ sin(𝑙𝑜𝑛2 − 𝑙𝑜𝑛1)/ sin(∆))

with θ being the azimuth and  the distance in degrees. To get azimuth (source to receiver) you
use event lat/lon as lat1/lon1; to calculate backazimuth (receiver to source) you use station
lat/lon as lat1/lon1. You will need these values in the next step.

Question/Task 3: Use the backazimuth to rotate the N-E components into radial and transverse
component. I.e. go from a geographic coordinate system to a coordinate system centered on the
great circle path.
rotaz ˽ 2,3 ˽ baz
with baz being the value of your determined backazimuth. Here and in the following ˽
indicates a space. Rotate the complete trace and remove the zoom window before rotation:
dtw
The result are two new traces which now represent radial (R) and transverse (T) component.
You can delete the original N, E components if you think you got the correct result.

del ˽ 2,3
Think about which surface wave (Rayleigh/Love) will appear on which components due to their
particle motion?
SOEE3260/5670M P4 - 4

Question/Task 4: Find the arrival time of the Rayleigh waves by looking at the particle motion:
a seismic phase arrival can be identified as a sudden change in the particle motion trajectory.
Roughly identify the earliest arrival of the Rayleigh wave and then check for retrograde
elliptical particle motion to identify the Rayleigh wave. To observe the particle motion diagram
in its own window run the command
open_xh3
that you have downloaded earlier. To take a look at the particle motion in a given time window
please use
pm/circle=0.02 ˽ 1,2 ˽ startwin endwin
startwin/endwin are the beginning and the end of the time window to calculate the particle
motion in. The first argument after the command shows the trace numbers of the two
components on which the Rayleigh wave should arrive.
Think about the particle motion of the Rayleigh wave and on which components of the
seismometer you would record these. Estimate from the traces where the Rayleigh wave might
arrive and use a 50 s time window with 50 s time steps. When you see the retrograde particle
motion, refine the time window until you are satisfied that you have found the correct start time.
You can define the start within a few seconds. From this arrival time you can calculate the
velocity of the earliest arrival of the Rayleigh wave by taking the distance and traveltime (use
your time window and the offset between earthquake origin and the start of the trace; 557
s). In SH:

Question/Task 5: Use a series of narrow bandpass filters to find the group velocity dispersion
curve of the Rayleigh wave. A range from 5 – 60 s with 5 s filter bandwidth (i.e. filters in the
range 5 s - 10 s, 10 s – 15 s, 15 s – 20 s, …) should work. In SeismicHandler you have to create
a filter first, then define it and apply it. To create a filter, use the command

makef
and follow the instructions on the screen. You want to create a bandpass (bp), using fft (f) and
define the period boundaries in seconds (s). Please use a third-order and define a filter name
such as bp_5s_10s_3 (for a third order bandpass with cutoffs of 5 s and 10 s).
SOEE3260/5670M P4 - 5

To define the filter, you have just created use the command
fili ˽ f ˽ bp_5s_10s_3
To filter the traces, use the command
filter ˽f ˽1,2

I.e.

SeismicHandler adds the filtered traces (of the trace 1,2 that you have requested) above the
original traces. It might be easier to identify the Rayleigh wave arrival in the enveloped trace,
which can be achieved using the command
envelope
I.e.

You should pick the traveltime at the beginning of the arriving wave package. Please remember
the time shift between the origin time and the start of the trace that you have to take into account
(= 557 s) to calculate the velocity. That information is given at the beginning. You can pick
traveltimes in SH using the command
time
which gives you a cursor to pick the time (time is given on the command line).
Use the letter e to finish picking times.
The high frequency group velocities (e.g. 5 – 10 s) might be difficult to pick. Keep in mind the
earliest arrival you were able to detect with the particle motion. No arrival should be earlier
than that. You can always calculate the particle motion if you want to check your arrival time.
Use the non-enveloped trace and proceed as described above.
Question/Task 7: Measure the velocities, using the information from traveltime and epicentral
distance, over a period range from 5 s to 60 s if possible. Add this information to a file called
dispersion_data.dat

with the first column containing the center of the period of the bandpass and the second column
the group velocity.

Plot the dispersion curve by plotting group velocity versus period for all filter ranges.

Question/Task 8: Try to find a velocity model along source to receiver by forward-modeling

a dispersion curve.
SOEE3260/5670M P4 - 6

Download the forward modeling tar-ball from Minerva. That tar-ball contains a script
(do_forward.csh) that calls 3 programs out of Hermann’s Computer programs for seismology
(http://www.eas.slu.edu/eqc/eqccps.html) (Herrmann, 2013). This script takes your velocity
model (e.g. my_model.mod) as input argument. I provide an example velocity model containing
a crust and mantle layer. You will need to change this velocity model to find a good model to
fit your data.

Look at the velocity model in a text editor. You can change the model by changing the lines
from Line 13 on. The format of the velocity model is shown in line 12. You only need to look
at the first 4 columns (Layer thickness, Vp, Vs, ). For our modeling the rest of the columns
are not important. You can add additional layers through additional lines. Each layer will have
a constant velocity.

The script also plots the modeled dispersion curve using the gmt (Wessel et al., 2013) script
plot_dispersion.gmt. It plots your own dispersion data from file dispersion_data.dat (on line 22
of the script). Please ask for help if that does not work.

Bibliography

Herrmann, R.B., 2013. Computer programs in seismology: An evolving tool for instruction and
research. Seismological Research Letters, 84, 1081-1088.

Stammler, K., 1993. SeismicHandler—Programmable multichannel data handler for interactive

and automatic processing of seismological analyses. Computers & geosciences, 19, 135-140.

Ward, J., Thorne, M., Nowacki, A. and Rost, S., 2021. Automatic slowness vector
measurements of seismic arrivals with uncertainty estimates using bootstrap sampling, array
methods and unsupervised learning. Geophysical Journal International, 226, 1847-1857.

Wessel, P., Smith, W.H., Scharroo, R., Luis, J. and Wobbe, F., 2013. Generic mapping tools:
improved version released. Eos, Transactions American Geophysical Union, 94, 409-410.

--- The End ---